Download Bose-Glass Phases of Ultracold Atoms due to Cavity Backaction Hessam Habibian,

PRL 110, 075304 (2013)
week ending
15 FEBRUARY 2013
PHYSICAL REVIEW LETTERS
Bose-Glass Phases of Ultracold Atoms due to Cavity Backaction
Hessam Habibian,1,2 André Winter,1 Simone Paganelli,2 Heiko Rieger,1 and Giovanna Morigi1
1
Theoretische Physik, Universität des Saarlandes, D-66123 Saarbrücken, Germany
Departament de Fı́sica, Universitat Autònoma de Barcelona, E-08193 Bellaterra, Spain
(Received 22 June 2012; revised manuscript received 16 November 2012; published 12 February 2013)
2
We determine the quantum ground-state properties of ultracold bosonic atoms interacting with the
mode of a high-finesse resonator. The atoms are confined by an external optical lattice, whose period is
incommensurate with the cavity mode wavelength, and are driven by a transverse laser, which is resonant
with the cavity mode. While for pointlike atoms photon scattering into the cavity is suppressed, for
sufficiently strong lasers quantum fluctuations can support the buildup of an intracavity field, which in
turn amplifies quantum fluctuations. The dynamics is described by a Bose-Hubbard model where the
coefficients due to the cavity field depend on the atomic density at all lattice sites. Quantum Monte Carlo
simulations and mean-field calculations show that, for large parameter regions, cavity backaction forces
the atoms into clusters with a checkerboard density distribution. Here, the ground state lacks superfluidity
and possesses finite compressibility, typical of a Bose glass. This system constitutes a novel setting where
quantum fluctuations give rise to effects usually associated with disorder.
DOI: 10.1103/PhysRevLett.110.075304
PACS numbers: 67.85.d, 03.75.Hh, 05.30.Jp, 32.80.Qk
Bragg diffraction is a manifestation of the wave properties of light and provides a criterion for the existence of
long-range order in the scattering medium [1]. Bragg
diffraction of light by atoms in optical lattices was measured for various geometries and settings, from gratings of
laser-cooled atoms [2–5] to ultracold bosons in the Mottinsulator (MI) phase [6]. In most of these setups, the
mechanical effect of the scattered light on the density
distribution of the atomic medium is negligible, while
photon recoil can give rise to visible effects in the spectrum
of the diffracted light [7].
Recent work proposed to use high-finesse optical resonators to enhance light scattering into one spatial direction
so as to reveal properties of the medium’s quantum state by
measuring the light at the cavity output [7,8]. These proposals assume that backaction of the cavity field on the
atoms can be discarded. Such an assumption is, however,
not valid in the regime considered in Refs. [9–14]: Here, the
strong coupling between cavity and atoms can induce the
formation of stable Bragg gratings in cold [9,10] and ultracold atomic gases [11–14] that coherently scatter light from
a transverse laser into the cavity mode. This phenomenon
occurs when the intensity of the pump exceeds a certain
threshold [10,11,13,15]. At ultralow temperatures the selforganized medium is a supersolid [13], while for larger
pump intensities incompressible phases are expected [16].
Let now the atoms be inside a high-finesse standingwave cavity and form a periodic structure, like the one
sketched in Fig. 1, whose period 0 =2 is incommensurate
with the cavity-mode wavelength . If the atoms are pointlike scatterers, or deep in a MI phase of the external
potential, there is no coherent scattering from a transverse
laser into the cavity mode [8,17]. Quantum fluctuations,
however, will induce scattering into the cavity, thus the
0031-9007=13=110(7)=075304(5)
creation of a weak potential with periodicity which is
incommensurate with 0 . In this Letter, we derive a BoseHubbard model for the system in Fig. 1 and show that
cavity backaction gives rise to an additional term in the
atomic energy which depends on the density at all lattice
sites. Even for weak probe fields this results in the formation of patterns which maximize scattering into the
cavity mode and can exhibit finite compressibility with
no long-range coherence. This feature, typical of disordered systems, corresponds to a Bose-glass phase for sufficiently deep potentials [18–21] and here emerges due to
the nonlocal quantum potential of the cavity field. Our
work extends studies predicting glassiness in multimode
cavities [22,23].
FIG. 1 (color online). Ultracold atoms are tightly confined by
an optical lattice of periodicity 0 =2. They are driven by a weak
transverse laser at Rabi frequency and strongly coupled to the
mode of a standing-wave cavity both at wavelength . Since and 0 are incommensurate, light is scattered into the cavity
mode because of quantum fluctuations. It then gives rise to an
incommensurate long-range interaction between the atoms
which modifies the density distribution.
075304-1
Ó 2013 American Physical Society
PRL 110, 075304 (2013)
week ending
15 FEBRUARY 2013
PHYSICAL REVIEW LETTERS
To analyze the quantum dynamics of the atom-cavity
system, we assume that the atomic motion is confined to
the x-z plane, where r ¼ ðx; zÞ is the atomic position. The
laser and cavity mode have wave vector kL ¼ kx^ and
kcav ¼ kz,
^ respectively, and polarizations such that only
a two-level dipolar transition at frequency !0 is driven.
Moreover, the scattering processes are coherent, which is
warranted when the detuning a ¼ !L !0 of the pump
frequency !L from !0 has modulus much larger than the
transition linewidth, the strength of the atom-photon coupling, and jc j, with c ¼ !L !c the detuning of the
pump from the mode frequency [13,16]. Let a^ and a^ y be
the operators annihilating and creating a cavity photon and
c^ ðrÞ be the atomic field operator with ½ c^ ðrÞ; c^ y ðr0 Þ ¼
ð2Þ ðr r0 Þ. In the reference frame rotating at frequency !L , the Hamiltonian governing the coherent dynamics reads H^ ¼ @c a^ y a^ þ H^ a þ H^ i , with H^ ‘¼a;i ¼
R
^ ðrÞ c^ ðrÞ. Here, H
^ ðrÞ ¼ @2 r2 þ V ðrÞ þ
dr c^ y ðrÞH
‘
a
0
2m
2
^ þ V1 cos ðkxÞ describes the atom dynamics in the
G s nðrÞ
absence of the resonator and in the presence of the external
periodic potential V0 ðrÞ ¼ Vt fcos 2 ðk0 zÞ þ cos 2 ðk0 xÞg
with wave number k0 , depth Vt along z, and aspect ratio
^ ¼ c^ y ðrÞ c^ ðrÞ is the atomic density
. Moreover, nðrÞ
operator, Gs is the strength of s-wave collisions, and V1 ¼
@2 =a such that jV1 j is the depth of the dynamical Stark
shift induced by the transverse standing-wave laser at Rabi
frequency . The atom-cavity coupling reads
^ ðrÞ ¼ @U cos2 ðkzÞa^ y a^ þ @S cos ðkxÞ cos ðkzÞða^ y þ aÞ;
^
H
i
0
0
(1)
where U0 ¼ g20 =a is the dynamical Stark shift per cavity
photon and g0 is the vacuum Rabi frequency at a cavityfield maximum [16,24]. The frequency S0 ¼ g0 =a is
the amplitude of scattering a laser photon into the cavity
mode by one atom. The corresponding term describes the
coherent pump on the cavity field via scattering by the
atoms and depends on the atomic positions within the field.
It gives rise to a large intracavity-photon number ncav when
the atoms form a Bragg grating with periodicity 2=k. This
would correspond to choosing k ¼ k0 : ncav then depends
on the balance between the superradiant scattering strength
S0 N and the rate of cavity loss . On the contrary, in this
work we take k and k0 to be incommensurate and analyze
the effect of cavity backaction when the atoms are tightly
confined by the potential V0 ðrÞ and weakly pumped by the
transverse laser.
We now sketch the derivation of the Bose-Hubbard
Hamiltonian for the atom dynamics. We first assume that
in the time scale t the atomic motion does not significantly evolveRwhile the cavity field has reached a local
^
steady state, tþt
aðÞd=t
a^ st , where the time evot
lution is governed by the
Heisenberg-Langevin
equation
pffiffiffiffiffiffi
^
^ H=i@
a^_ ¼ ½a;
a^ þ 2a^ in ðtÞ, with a^ in ðtÞ the input
noise, ½a^ in ðtÞ; a^ yin ðt0 Þ ¼ ðt t0 Þ. The conditions for a
time-scale separation are set by the inequality jc þ
ijt 1 and by the assumption that coupling strengths
between atoms and fields are much smaller than 1=t [25].
The stationary field reads
pffiffiffiffiffiffi
^
S0 Z
i 2a^ in
a^ st ¼
þ
; (2)
^ þ i ð U YÞ
^ þ i
ð U YÞ
c
0
c
0
^ ¼
with a^ in the input noise averaged over t. Here, Y
R 2
R
^ ¼ d2 r cos ðkzÞ cos ðkxÞnðrÞ
^
^
d rcos 2 ðkzÞnðrÞ
and Z
describe the shift of the cavity resonance and coherent
scattering amplitude, respectively, due to the atomic density distributions. The noise term can be neglected when
the mean intracavity-photon number is larger than its
^ . In this limit, one can monifluctuations, i.e., jS0 hZij
tor the p
state
ffiffiffiffiffiffi of the atoms via the field at the cavity output,
a^ out ¼ 2a^ st a^ in [17]. Using Eq. (2) for a^ in Eq. (1)
leads to an effective Hamiltonian, from which we derive a
Bose-Hubbard model assuming that the atoms are tightly
trapped in the lowest band of V0 ðrÞ. We use c^ ðrÞ ¼
P
^
i wi ðrÞbi , with wi ðrÞ the Wannier function centered at
the minimum ri of potential V0 ðrÞ and b^i the bosonic
operator annihilating a particlepatffiffiffiffiri . We apply the thermodynamic limit where S0 ¼ s0 = K and U0 ¼ u0 =K with K
the number of lattice sites [16] and obtain the BoseHubbard Hamiltonian
K X
X y
U
^
^
^
^
H BH ¼
n^ i ðn^ i 1Þ ^ i n^ i ti ðbi bj þ H:c:Þ ; (3)
i¼1 2
hj;ii
^y ^
with
R 2n^ i ¼ bi b4i the on site atomic density, U ¼
Gs d r½wi ðrÞ the on site interaction strength, ^ i the
site-dependent chemical potential, t^i the tunneling coefficient, and hj; ii the sum over nearest neighbors. Here, ^ i ¼
ð0Þ þ ^ i , t^i ¼ tð0Þ þ t^i , where ð0Þ ¼ E0 Vt X0
and tð0Þ ¼ E1 VtRX1 are constant over the lattice, with
E
¼ @2 =ð2mÞ d2 rwi ðrÞr2 wiþs ðrÞ and Xs¼0;1 ¼
Rs¼0;1
2
d rfcos2 ðk0 zÞ þ cos 2 ðk0 xÞgwi ðrÞwiþs ðrÞ. The terms
^ i and t^i are due to the pump and cavity incommensurate potentials and vanish when the pump laser is off,
¼ 0. In this limit the model reduces to the regular
Bose-Hubbard model exhibiting a MI-superfluid (SF) transition as a function of the parameters and t [18]. When
0, still ti ¼ tð0Þ since t^i is negligible, while
^ i ¼ V1 J0ðiÞ @
s20
^ ^ eff ZðiÞ þ u0 Y
^ ðiÞ Þ:
ð2
0
0
2
2
^
eff þ (4)
R
Here, J0ðiÞ ¼ d2 rcos 2 ðkxÞwi ðrÞ2 scales the strength of the
classical transverse potential due to the laser and is constant for the sites with the same x value, Z0ðiÞ ¼
R
R 2
d rcosðkxÞcosðkzÞwi ðrÞ2 and Y0ðiÞ ¼ d2 rcos 2 ðkzÞ
wi ðrÞ2 are due the cavity field, whereas ^ eff ¼ c P
u0 i Y0ðiÞ n^ i =K is an operator and accounts for the frequency
075304-2
densities n ¼ 1, 2, and the blue regions a compressible
phase where the SF density vanishes, while outside the
phase is SF. The effect of cavity backaction is evident at
^ 0: Here the size of the MI
low tunneling, where hi
regions is reduced. At larger tunneling a direct MI-SF
transition occurs, and the MI-SF phase boundary merges
with the one found for s0 ¼ 0: In fact, for larger quantum
^ ! 0 in the thermodynamic limit. This
fluctuations hi
feature is strikingly different from the situation in which
the incommensurate potential is classical [30,31]: There,
the MI lobes shrink at all values of t with respect to the
pure case. The atomic density and corresponding value of
^ are displayed in Figs. 2(b) and 2(c) as a function of hi
for different values of the transverse laser intensity (thus
s0 ): The incommensurate potential builds in the blue
region of the diagram, which we label by BG, where the
number of intracavity photons does not vanish. Figure 2(d)
displays the local density distribution hn^ i i and the local
density fluctuations in the BG region: The density oscillates in a quasiperiodic way over clusters in which the
atoms scatter in phase into the cavity mode. The fluctuations are larger at the points where ZðiÞ
0 , which oscillates at
the cavity mode wavelength, becomes out of phase with the
trapping potential. Note that by means of this rearrangement the fields emitted by each clusters add up coherently:
In this way the density distribution maximizes scattering
into the cavity mode. This distribution is reminiscent of a
density wave, which is characterized by a zero order
parameter and nonvanishing compressibility. We denote
the corresponding region by BG, which stands for Bose
glass, using the terminology applied in Refs. [30,31] to
similar density distributions found in the bichromatic
Bose-Hubbard model.
shift due to the atoms [24,26]. Remarkably, the cavity
effects are scaled by the operator
^ ¼
K
X
^ i =K;
ZðiÞ
0 n
(5)
i¼1
which originates from the long-range interaction mediated
by the cavity field and is related to the mean intracavity
^ 2 i. Its mean value vanishes
photon number since ncav / h
^ MI / P ZðiÞ ¼ 0,
when the atomic gas forms a MI state: hi
i 0
since there is no coherent scattering into the cavity mode.
^ SF ! 0. Close to the MI-SF
Also deep in the SF phase hi
phase transition, however, fluctuations in the atomic den^ and, hence, to a finite
sity lead to finite values of hi
intracavity photon number. The dependence of the chemi^ is a peculiar property of our
cal potential on the operator model that distinguishes it from the case of a bichromatic
optical lattice [20,21], where the strength of the incommensurate potential is independent of the phase of the
ultracold atomic gas.
We first analyze the ground state of the Bose-Hubbard
Hamiltonian, Eq. (3), when the system can be reduced
to a one-dimensional (1D) lattice along the cavity axis
( 1). In 1D the effect of the transverse potential [first
term in Eq. (4)] is a constant shift which can be reabsorbed
in the chemical potential. Figure 2(a) displays the phase
diagram in the -t plane evaluated by means of a quantum
Monte Carlo calculation [27]. The compressibility
is deterP
@n
mined by using ¼ @
, with n ¼ i hn^ i i=K the mean
density, while the SF density is obtained by extrapolating
the Fourier transform of the pseudo current-current correlation function Jð!Þ [27] to zero frequency [see the inset in
Fig. 2(b)]. The gray regions indicate the MI states at
2
1.2
2
1
1.5
1.5
4
0.8
3
1
2
1
0.5
0.6
1
0
-80
0.5
0
week ending
15 FEBRUARY 2013
PHYSICAL REVIEW LETTERS
PRL 110, 075304 (2013)
0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
0
0.3
0.2
0.1
0
-0.2
-40
0
40
80
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0
0
10
20
30
40
50
60
70
80
90 100
FIG. 2 (color online). Results of quantum Monte Carlo simulations for a 1D lattice ( 1) with 100 sites and periodic
boundary conditions. (a) Phase diagram in the -t plane for s0 ¼ 0:004. The gray regions indicate incompressible phases at density
n ¼ 1, 2, the blue region the gapless phases with vanishing SF density. The dotted lines indicate the incompressible phases in the
^ versus for t ¼ 0:053U and s0 = ¼ 0:003 (triangles),
absence of the pump laser (s0 ¼ 0) [27]. (b) Linear density n and (c) hi
s0 = ¼ 0:004 (circles), and s0 ¼ 0 (squares): The number of photons is different from zero for the parameters of the BG regions in (a).
Inset: Fourier transform of the pseudo current-current correlation function Jð!Þ [27] for the parameters indicated by the arrows in (b).
hn^ 2i i hn^ i i2 (filled points
(d) Local density distribution hn^ i i (empty points joined by the blue curve) and local density fluctuations
pffiffiffiffi
joined by the red curve) as a function of the site for ¼ 0 and s0 ¼ 0:004. The values of s0 ¼ K g0 =a are found from g0 =2 ¼
14:1 MHz, =2 ¼ 1:3 MHz, and detunings about a =2 ¼ 58 GHz [13]. The other parameters are c ¼ 5, k=k0 ¼ 785=830 as
in Ref. [13], and G s was taken from Ref. [37].
075304-3
PRL 110, 075304 (2013)
PHYSICAL REVIEW LETTERS
70
week ending
15 FEBRUARY 2013
1
Lattice sites’ labels along x
60
60
40
50
20
0.8
20
40
40
0.6
60
30
0.4
20
0.2
(b)
10
10
20
30
40
50
60
70
0
Lattice sites’ labels along z
FIG. 3 (color online). Results of a mean-field calculation for a 2D lattice with 70 70 sites and periodic boundary conditions.
(a) Order parameter in the -t plane when s0 ¼ 0:15 (c ¼ 5). Inset: Phase diagram when cavity backaction is negligible
(c ¼ 300). The dotted and solid curves delineate the regions with a small, arbitrarily chosen threshold (here 0.02) for and the
density fluctuations, respectively, while inside the dashed curve the number of photons is at least 100 times larger than outside. The
localized regions at smaller deep in the SF phase are due to the transverse incommensurate pump. (b) Local density distribution hn^ i i
at ¼ 0:156U and t ¼ 0:01U [point in (a) at n ¼ 0:64]. Inset: Corresponding density distribution in the -t plane for c ¼ 300
(here n ¼ 0:625). (c) Photon emission rate at the cavity output, nout ¼ hayout aout i, as a function of t for n ¼ 0:64 and in units of
ðmax Þ
ðmax Þ
¼ s20 K=ð^ 2eff þ 2 Þ is the intracavity photon number when all
nð0Þ
out ¼ ncav . Inset: Corresponding order parameter. Here, ncav
atoms scatter in phase into the cavity mode, s0 ¼ 0:15 is consistent with the parameters of Ref. [13] for 300 300 sites, and G s was
taken from Ref. [38].
The two-dimensional (2D) case is studied by means of a
mean-field calculation [18,32] for a pumping strength s0
such that max fjh^ i ijg < U. Here, the transverse laser
determines the transverse density distribution even when
the cavity field is zero. In this case the MI lobes at t ¼ 0
shrink due to the classical incommensurate field. Because of
the classical potential, in general no direct MI-SF transition
is expected [33]. Figure 3(a) displays the order parameter,
P
¼ i hb^i i=K, in the -t plane and for density n 1.
The solid curve indicates where the gap in the spectrum
is different from zero, corresponding to vanishing density
P
fluctuations % ¼ ðn2 n 2 Þ1=2 , for n2 ¼ i hn^ 2i i=K
[34,35]. The dashed line separates the parameter region
where the number of intracavity photons is at least 2 orders
of magnitude larger than outside. For comparison the diagram for ncav ’ 0 is reported in the inset. A larger region
with vanishing appears when the atoms are strongly
coupled to the cavity. The analysis of the corresponding
density distribution shows a peculiar behavior. Figure 3(b)
displays the local density for c ¼ 5 for the parameters
indicated by the point in (a): The coupling with the resonator induces the formation of clusters with checkerboard
density distribution, where the atoms scatter in phase into
the cavity mode. At the border of the clusters, the density
fluctuates, so as to allow the fields scattered by each cluster
to interfere constructively. The inset shows that the clustering disappears when cavity backaction is negligible. In this
latter case the field at the cavity output is zero, while when
the effect of cavity backaction is relevant it has a finite
intensity for a finite range of values of t [Fig. 3(c)]. When
the tunneling rate is instead sufficiently large, there is no
clustering and the atomic density becomes uniform along
the cavity axis. Thus, the competition between the mechanical effects of the cavity field and quantum fluctuations leads
to the creation of these clusters, which exhibit properties
ranging from a BG to a strongly correlated SF and then
disappear when tunneling becomes large.
In conclusion, we predicted an instance of selforganization of a quantum gas in a high-finesse resonator
that is triggered by quantum fluctuations of the atomic
motion and could be observed in existing experimental
setups [13,36].
We acknowledge discussions with A. Niederle, G. De
Chiara, S. Fernández-Vidal, M. Lewenstein, and H. Ritsch
and support by the European Commission (IP AQUTE), the
Spanish Ministerio de Ciencia y Innovación (QOIT:
Consolider-Ingenio 2010; FPI; Juan de la Cierva), the
Generalitat de Catalunya (SGR2009:00343), and the
German Research Foundation.
[1] R. Lifshitz, Z. Kristallogr. 222, 313 (2007).
[2] G. Birkl, M. Gatzke, I. H. Deutsch, S. L. Rolston, and
W. D. Phillips, Phys. Rev. Lett. 75, 2823 (1995).
[3] M. Weidemüller, A. Hemmerich, A. Görlitz, T. Esslinger,
and T. W. Hänsch, Phys. Rev. Lett. 75, 4583 (1995); M.
Weidemüller, A. Görlitz, T. W. Hänsch, and A.
Hemmerich, Phys. Rev. A 58, 4647 (1998).
[4] S. Slama, C. von Cube, B. Deh, A. Ludewig, C.
Zimmermann, and Ph. W. Courteille, Phys. Rev. Lett. 94,
193901 (2005).
[5] S. Slama, C. von Cube, M. Kohler, C. Zimmermann, and
Ph. W. Courteille, Phys. Rev. A 73, 023424 (2006).
[6] C. Weitenberg, P. Schauß, T. Fukuhara, M. Cheneau, M.
Endres, I. Bloch, and S. Kuhr, Phys. Rev. Lett. 106,
215301 (2011).
[7] S. Rist, C. Menotti, and G. Morigi, Phys. Rev. A 81,
013404 (2010); J. S. Douglas and K. Burnett, Phys. Rev.
A 84, 033637 (2011).
075304-4
PRL 110, 075304 (2013)
PHYSICAL REVIEW LETTERS
[8] I. B. Mekhov, C. Maschler, and H. Ritsch, Nat. Phys. 3,
319 (2007); Phys. Rev. Lett. 98, 100402 (2007).
[9] P. Domokos and H. Ritsch, Phys. Rev. Lett. 89, 253003
(2002).
[10] A. T. Black, H. W. Chan, and V. Vuletić, Phys. Rev. Lett.
91, 203001 (2003).
[11] D. Nagy, G. Szirmai, and P. Domokos, Eur. Phys. J. D 48,
127 (2008).
[12] J. Keeling, M. J. Bhaseen, and B. D. Simons, Phys. Rev.
Lett. 105, 043001 (2010).
[13] K. Baumann, C. Guerlin, F. Brennecke, and T. Esslinger,
Nature (London) 464, 1301 (2010).
[14] K. Baumann, R. Mottl, F. Brennecke, and T. Esslinger,
Phys. Rev. Lett. 107, 140402 (2011).
[15] J. K. Asbóth, P. Domokos, H. Ritsch, and A. Vukics, Phys.
Rev. A 72, 053417 (2005).
[16] S. Fernández-Vidal, G. De Chiara, J. Larson, and
G. Morigi, Phys. Rev. A 81, 043407 (2010).
[17] H. Habibian, S. Zippilli, and G. Morigi, Phys. Rev. A 84,
033829 (2011).
[18] M. P. A. Fisher, P. B. Weichman, G. Grinstein, and
D. S. Fisher, Phys. Rev. B 40, 546 (1989).
[19] R. T. Scalettar, G. G. Batrouni, and G. T. Zimanyi, Phys.
Rev. Lett. 66, 3144 (1991); W. Krauth, N. Trivedi, and
D. Ceperley, Phys. Rev. Lett. 67, 2307 (1991).
[20] B. Damski, J. Zakrzewski, L. Santos, P. Zoller, and
M. Lewenstein, Phys. Rev. Lett. 91, 080403 (2003).
[21] B. Deissler, M. Zaccanti, G. Roati, C. D’Errico, M.
Fattori, M. Modugno, G. Modugno, and M. Inguscio,
Nat. Phys. 6, 354 (2010).
[22] S. Gopalakrishnan, B. L. Lev, and P. M. Goldbart, Nat.
Phys. 5, 845 (2009).
[23] S. Gopalakrishnan, B. L. Lev, and P. M. Goldbart, Phys.
Rev. Lett. 107, 277201 (2011); P. Strack and S. Sachdev,
Phys. Rev. Lett. 107, 277202 (2011).
[24] C. Maschler and H. Ritsch, Phys. Rev. Lett. 95, 260401
(2005).
week ending
15 FEBRUARY 2013
[25] At finite temperature T, B T @=t; see Ref. [16].
[26] S. Gupta, K. L. Moore, K. W. Murch, and D. M. StamperKurn, Phys. Rev. Lett. 99, 213601 (2007).
[27] We use a grand-canonical realization of the quantum
Monte Carlo simulation [28], with which we identify the
discontinuity of n at the border of the BG phase [Fig. 2(b)].
A conventional canonical simulation is used at the
MI-SF border [29]. The blue points in Fig. 2(a) are
determined when the jumps in the plots of n versus occur.
[28] P. Niyaz, R. T. Scalettar, C. Y. Fong, and G. G. Batrouni,
Phys. Rev. B 50, 362 (1994).
[29] G. G. Batrouni, R. T. Scalettar, and G. T. Zimanyi, Phys.
Rev. Lett. 65, 1765 (1990); G. G. Batrouni and R. T.
Scalettar, Phys. Rev. B 46, 9051 (1992).
[30] X. Deng, R. Citro, A. Minguzzi, and E. Orignac, Phys.
Rev. A 78, 013625 (2008).
[31] G. Roux, T. Barthel, I. P. McCulloch, C. Kollath, U.
Schollwöck, and T. Giamarchi, Phys. Rev. A 78, 023628
(2008).
[32] K. Sheshadri, H. R. Krishnamurthy, R. Pandit, and T. V.
Ramakrishnan, Europhys. Lett. 22, 257 (1993).
[33] L. Pollet, N. V. Prokof’ev, B. V. Svistunov, and M. Troyer,
Phys. Rev. Lett. 103, 140402 (2009).
[34] B. Damski and J. Zakrzewski, Phys. Rev. A 74, 043609
(2006).
[35] M. Ła̧cki, S. Paganelli, V. Ahufinger, A. Sanpera,
and J. Zakrzewski, Phys. Rev. A 83, 013605 (2011);
S. Paganelli, M. Ła̧cki, V. Ahufinger, J. Zakrzewski,
and A. Sanpera, J. Low Temp. Phys. 165, 227
(2011).
[36] K. J. Arnold, M. P. Baden, and M. D. Barrett, Phys. Rev.
Lett. 109, 153002 (2012).
[37] T. Volz, S. Dürr, S. Ernst, A. Marte, and G. Rempe, Phys.
Rev. A 68, 010702(R) (2003).
[38] P. Krüger, Z. Hadzibabic, and J. Dalibard, Phys. Rev. Lett.
99, 040402 (2007).
075304-5